Indonesian Journal of Electrical Engineering and Informatics (IJEEI)
Not a member yet
776 research outputs found
Sort by
Multi-Objective Reinforcement Learning Based Algorithm for Dynamic Workflow Scheduling in Cloud Computing
Full text linkIt is essential to consider the infrastructures of workflows as a critical research area where even slight optimizations can significantly impact infrastructure efficiency and the services provided to users. Traditional workflow scheduling approaches using heuristics may not be efficient due to the dynamic workloads and diverse resources of cloud infrastructure. Additionally, the resources at any given time have different states that must be considered during workflow scheduling. The emergence of artificial intelligence has made it possible to address the dynamics and diverse resources of cloud computing during workflow management. In particular, reinforcement learning enables understanding the environment at runtime with an actor and critic approach to make well-informed decisions. Our paper introduces an algorithm called Multi-Objective Reinforcement Learning based Workflow Scheduling (MORL-WS). Our empirical study with various workflows has demonstrated that the proposed multi-objective reinforcement learning-based approach outperforms many existing scheduling methods, especially regarding makespan and energy efficiency. The proposed method with the Montage workflow demonstrated superior performance compared to scheduling 1000 tasks, achieving a least makespan of 709.26 and least energy consumption of 72.11 watts. This indicates that the proposed method is suitable for real-time workflow scheduling applications
Performance Evaluation of Advanced PLL Techniques For Accurate FFPS Component Extraction
Full text linkIt is very necessary to adopt fundamental frequency positive sequence (FFPS) element extraction methods in order to maximise the efficiency of integrating and handling the use of renewable energy sources (RES). The decision to act in this manner is made with the purpose of contributing to the accomplishment of the aforementioned aim. The capability of the synchronous references frame phase-locked loop to reject variations over a broad variety of grid conditions is enhanced as a result of this. This is particularly true for voltage sags and surges that are accompanied by harmonics, irrespective of whether the harmonics are the result of balanced or unbalanced electrical current fluctuations. The accuracy of the extraction of FFPS components is significantly influenced by the frequency deviation in SRF-PLL systems. The frequency deviation is another critical component. This is a result of the frequency deviation not remaining constant. An investigation is being conducted to ascertain the effectiveness of a various advanced PLL techniques, such as the cascaded delayed signal cancellation (CDSC), the dual second-order generalized integrator (DSOGI) and the multiple delayed signal cancellation (MDSC). The objective of conducting this assessment is to facilitate the evaluation of the efficacy of these strategies, which is the reason for its implementation. The CDSC and MDSC PLL have been demonstrated to be preferable to other PLLs due to their ability to distinguish between even and odd harmonics. This is due to the fact that each of these harmonics possesses its own distinctive characteristics. This may be attributable to its capacity to independently identify either harmonic. The MATLAB simulation results is provided to demonstrate the exceptional performance of these highly advanced PLLs
Photometric Stereo-based Woven Fabric Pattern Recognition Using Wavelet Image Scattering
Full text linkThe weave pattern is a crucial factor that enhances the strength and stability of the fabric. Pattern recognition of woven fabric based on vision methods has been widely developed. In this research, woven fabric's basic weaving pattern recognition is based on photometric stereo images. First, six images of woven fabric were taken, each with a different direction of light. Next, an unbiased photometric stereo algorithm was used to reconstruct the six images. This paper used 23 grayscale photometric stereo images measuring 400 x 300 pixels. Augmentation techniques were carried out to produce 458 images consisting of 240 plain woven images, 159 twill woven images, and 60 satin woven images. The training data set consists of 367 images, and testing consists of 192 images. The feature extraction method uses wavelet image scattering and classification using Principal Component Analysis (PCA) and Support Vector Machine (SVM). The wavelet image scattering method effectively extracts texture features of photometric stereo images of diverse woven fabrics, while the PCA and SVM methods successfully classify the basic woven fabric patterns. The results of recognizing the basic woven fabric pattern using PCA and SVM classification obtained an accuracy of 98.57%
Implementation of Image Processing and CNN for Roasted-Coffee Level Classification
Full text linkThe roasting process of coffee beans plays a crucial role in the development of chemicals responsible for the rich color and complex flavors characteristic of well-roasted coffee. One approach to understanding this process involves assessing the roast level, which varies in color from light to dark, with intermediate levels in between. In this study, image processing was performed using Convolutional Neural Networks (CNNs), a widely used method for image classification. The objective was to utilize the LAB color model and the CNN framework to classify the roast levels of coffee beans based on images from files or video streams. The study also details the hardware and software tools employed. A user-friendly graphical interface was developed to ensure ease of use, requiring minimal training for efficient operation. The research successfully designed, developed, and implemented an application for classifying coffee bean roast levels using two methods: LAB color model image processing and the CNN model. Consequently, the system can recognize roast levels based on the outputs from both the LAB model and the CNN model. This research represents a preliminary effort and requires further development to support more extensive studies. Ultimately, it serves as a foundation for future exploration and the application of embedded system-based solutions for assessing coffee bean maturity levels in alignment with Agtron classification standards
The Use of Green-Phosphor LuAG:Ce-Al2O3 for HighLuminosity Light Emitting Diode Packages
Full text linkThe LuAG:Ce-Al2O3 (LAGCA) green phosphor ceramic (GPC) is proposed for high-power white light emitting diodes (LEDs) in this paper. The luminescent properties of the GPC are examined with proper characterizing tools and under laser excitation. Then, LAGCA ceramic layer of 0.6 thickness is applied to fabricate the white LED. The results show LAGCA GPC is promising for high-power LED applications. The phosphor ceramic presents high thermostability and quantum efficacy and intense green emission peaking at nearly 550 nm. In the LED package, the amount of LAGCA in the composite layer is varied. The increasing dosage of LAGCA gives enhancement to the lumen output of the LED. However, the correlated color temperature stability and chromatic rendition declines. Thus, further improvements in LAGCA ceramic need to be carried out in the future works. Besdies, with the intense green emission, the LAGCA ceramics can be combined with red luminescent materials to increase the color performance of the LED lighting
Classification of Human Emotions Based on Javanese Speech Using Convolutional Neural Network and Multilayer Perceptron
Full text linkEmotions in speech are considered a basic principle of human interaction and play an important role in decision-making, learning, and everyday communication. Research on speech emotion recognition is still being carried out by many researchers to develop speech emotion recognition models with better performance. In this research, we combine the application of data augmentation techniques (Add Noise, Time Stretch, and Pitch Shift) to increase the data size of the Javanese Speech Emotion Database (Java-SED). Mel Frequency Cepstral Coefficients (MFCC) is used for feature extraction and then builds a Convolutional Neural Network (CNN) model and applies Multilayer Perceptron (MLP) to classify human emotions from sound. In this research, we produced eight experimental models with a combination of different augmentation techniques. The CNN model parameters include 40 input neurons, four hidden layers with varying neuron counts, Relu activation functions, L2 regularization, dropout rates, Adam optimization, and ModelCheckpoint callbacks to save the best model based on validation loss. From the results of the evaluation that has been carried out, the CNN algorithm produces the highest performance with an accuracy of 96.43%, recall of 96.43%, precision of 96.57%, F1-score of 96.48%, and kappa of 95.71% by applying the Add Noise technique, Time Stretch, and Pitch Shift
Driver Drowsiness Detection using Hybrid Algorithm
Full text linkIn this work we focus on the discernment of sleepiness in drivers’ drowsiness proposing a hybrid algorithm which aims to confirm whether the driver's level of attention has decreased owing to a nap or any other medical issue, such as brain problems. Therefore, the proposed hybrid algorithm uses both Haarcascade classifier and Convolutional Neural Network (CNN) algorithm to detect drivers’ drowsiness. The driver's eyes will be monitored and an alert sound will be generated by Raspberry Pi module, but the face must be moving in real time, and the aspect ratio must be between 16:9 and 1.85:1. People often feel sleepy since activities like driving call for a proper mental state, and bad work-life balance has additional negative repercussions. When we give input through normal camera it analyses drivers state of eyes and mouth, actually it checks aspect ratio of eye. We proved in comparative trials that our hybrid algorithm beats current driving fatigue detection algorithms in speed as well as accuracy
Performance Enhancement of Decode and Forward Relaying Network in a Log- normal Fading Channel using Diversity Technique
Full text linkThe demand for wireless communication services is increasing daily due to several emerging applications of wireless communication system. However, the services provided by wireless communication is affected by obstruction along the path of propagation resulting in scattering of signal at the receiver. Decode and Forward (DF) relaying network used in addressing the problem also suffer from signal outage at the destination due to inability of relay to decode the transmitted signal at the relay node. Hence, in this paper, performance enhancement of DF relaying network is proposed using Time Diversity (TD) at the source with hybrid Threshold Combiner and Equal Gain Combiner (TC-EGC) at the destination. The various copies of the transmitted signal are received at the DF relay node to carry out relay selection by selecting relay with signal strength greater than the set threshold of 3 dB. The selected relays decode and re-encode the received signal before been propagated to the destination. The various copies of the signal received at the destination with varying paths ‘L’ (2, 3 and 4) are combined using TC-EGC. Mathematical expressions of Outage Probability (OP) and Bit Error Rate (BER) for the proposed technique are derived using Probability Density Function (PDF) of the signal received. The proposed DF technique is simulated using MATLAB R2018a and validated using OP and BER by comparing with the conventional DF cooperative relaying network. The proposed technique improved the performance of conventional DF cooperative relaying network with reduced BER and OP
A 76 GHz Millimeter-Wave Marine Radar Antenna Design
Full text linkIn this work, a 76 GHz microstrip antenna array is proposed for usage in the mm-wave Marine radar application. Millimeter-wave radars are commonly used in automotive applications and a lot of effort is made in this way but it is also can be used in marine applications as they work robustly in bad weather conditions such as fog, dust, smog, smoke, and water vapor. So, it will be very helpful in marine applications. The proposed array antenna is a corporate Series feed 24×8 antenna array that has achieved a return loss of -26.4 dB, a gain of 23.5dBi, bandwidth of 5.2 GHz, and sidelobe levels of -21.4 dB in Hplane and -14 dB in E-plane. This antenna array's 3dB angular width equals 10.9 ° in the H-plane and 5.9 ° in the E-plane. That makes it a suitable choice for the mm-wave marine radar antenna. The final design of the antenna is acceptable compared with another previous work, making this design more considerable as will be shown. Also, an antenna array with 3 transmitters and 4 receivers is presented. Each antenna is a 24-element. the Dolph-Chebyshev technique is utilized to taper the patches. The antenna has been manufactured, and the results of the simulation are confirmed by the experimental measurements
Classifications of Arabic Customer Reviews Using Stemming and Deep Learning
Full text linkWith the emergence of AI text-based tools and applications, the need to present and investigate text-processing tools has also been raised. NLP tools and techniques have developed rapidly for some languages, such as English. However, other languages, like Arabic, still need to present more methods and techniques to present more explanations. In this study, we present a model to classify customer reviews which are written in Arabic. The HARD dataset is used to be adopted as the dataset. Three Deep Learning classifiers are adopted (CNN, LSTM, RNN). In addition to that, three stemmers are used as text processing techniques (Khoja, Snowball, Tashaphyne). Furthermore, another three feature extraction methods were utilized (TF-IDF, N-gram, BoW). The results of the model presented several explanations. The best performance resulted from using (CNN+ Snowball+ N-Gram) with an accuracy of (%93.5). The results of the model stated that some classifiers are sensitive toward using different stemmers, also some accuracy performance can be affected if there are different feature extraction methods used. Either stemming of feature extraction has an impact on the accuracy performance. The model also proved that the dialectal language could cause some limitations since different dialects can give conflict meaning across different regions or countries. The outcomes of the study open the gate towards investigating other tools and methods to enrich Arabic natural language processing and contribute to developing new applications that support Arabic content